Fibrillin-1-regulated miR-122 has a critical role in thoracic aortic aneurysm formation

Abstract

Thoracic aortic aneurysms (TAA) in Marfan syndrome, caused by fibrillin-1 mutations, are characterized by elevated cytokines and fragmentated elastic laminae in the aortic wall. This study explored whether and how specific fibrillin-1-regulated miRNAs mediate inflammatory cytokine expression and elastic laminae degradation in TAA. miRNA expression profiling at early and late TAA stages using a severe Marfan mouse model (Fbn1^mgR/mgR) revealed a spectrum of differentially regulated miRNAs. Bioinformatic analyses predicted the involvement of these miRNAs in inflammatory and extracellular matrix-related pathways. We demonstrate that upregulation of pro-inflammatory cytokines and matrix metalloproteinases is a common characteristic of mouse and human TAA tissues. miR-122, the most downregulated miRNA in the aortae of 10-week-old Fbn1^mgR/mgR mice, post-transcriptionally upregulated CCL2, IL-1β and MMP12. Similar data were obtained at 70 weeks of age using Fbn1^C1041G/+ mice. Deficient fibrillin-1–smooth muscle cell interaction suppressed miR-122 levels. The marker for tissue hypoxia HIF-1α was upregulated in the aortic wall of Fbn1^mgR/mgR mice, and miR-122 was reduced under hypoxic conditions in cell and organ cultures. Reduced miR-122 was partially rescued by HIF-1α inhibitors, digoxin and 2-methoxyestradiol in aortic smooth muscle cells. Digoxin-treated Fbn1^mgR/mgR mice demonstrated elevated miR-122 and suppressed CCL2 and MMP12 levels in the ascending aortae, with reduced elastin fragmentation and aortic dilation. In summary, this study demonstrates that miR-122 in the aortic wall inhibits inflammatory responses and matrix remodeling, which is suppressed by deficient fibrillin-1–cell interaction and hypoxia in TAA.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-022-04337-8.

Introduction

Thoracic aortic aneurysm (TAA) is one of the most fatal vascular disease, characterized by fragmentation of the elastic laminae, proteoglycan accumulation, and loss of smooth muscle cells in the ascending aortic wall [1]. However, the molecular mechanisms contributing to TAA pathogenesis are still insufficiently explored. Single gene mutations cause up to 20% of TAAs [2]. Marfan syndrome (MFS, OMIM# 154700) is an autosomal dominantly inherited connective tissue disorder caused by heterozygous mutations in the fibrillin-1 gene [3]. TAA develops in 85–90% of MFS patients [4], which makes MFS the most prominent disorder among the genetic syndromes that predispose to TAA. Today, more than 1800 mutations are known in fibrillin-1 that cause MFS [5], with an estimated prevalence of 2–3 in 10,000 individuals [6].

Fibrillin-1 is a 350 kDa glycoprotein that is secreted and assembled in the extracellular matrix of connective tissues, including aorta, skin, bone, and eye. It comprises the core of fibrillin-containing microfibrils, which provide the principal scaffold for elastic fiber and lamellae formation required for elastic tissue properties. Fibrillin-1 microfibrils in the aortic wall connect the smooth muscle cells to elastic laminae [7]. Disrupted smooth muscle cell mechanosensing caused by fibrillin-1 mutations is one of the drivers of TAA formation in MFS (for review see Milewicz et al. [8]). This view is supported by the fact that mutations in proteins controlling smooth muscle cell contractility, including MYH11 [9], ACTA2 [10], and TES [11], also lead to TAA formation. Fibrillin-1 microfibrils and its associated proteins sequester transforming growth factor-β (TGF-β), an important regulatory cytokine in TAA pathogenesis. Deficiency of functional microfibrils leads to upregulation of TGF-β in both human and mouse MFS [12, 13].

Inflammatory cell infiltration into the aortic adventitia and elastic lamina fragmentation was observed in aortic specimen from human TAA patients, as well as from Fbn1^mgR/mgR and Fbn1^C1041G/+ mouse models of MFS [14–16]. Pro-inflammatory cytokines, including CCL2 [17], IL-1β [18], and IL-6 [17] are upregulated in the aortic wall of TAA mouse models or human patients. Among them, IL-1β can induce the expression of CCL2 and multiple matrix metalloproteinases (MMP) from smooth muscle cells, including MMP1, 2, 3 and 9 [19]. Elevated MMP2 and MMP9 are associated with smooth muscle phenotypic changes and elevated TGF-β signaling at the aneurysmal lesions in MFS patients [20], and in Fbn1^mgR/mgR and Fbn1^C1041G/+ MFS mice [21, 22]. Inhibition of MMP2 and MMP9 by doxycycline ameliorates TAA progression in Fbn1^mgR/mgR mice consistent with reduced TGF-β signaling and decreased elastic lamina fragmentation [21]. Thus, it is important to understand how pro-inflammatory cytokines and MMPs are regulated during TAA pathogenesis.

Oxidative stress with excessive reactive oxygen species is another hallmark reported in aneurysmal tissues of TAA patients, including MFS [23]. Excessive reactive oxygen species can stabilize hypoxia-inducible factor 1α (HIF-1α) under hypoxic conditions [24]. HIF-1α is a master regulator of oxygen homeostasis, protecting cells under hypoxic conditions [25], observed in both TAA and abdominal aortic aneurysms (AAA) in human patients [26, 27]. Inhibition of HIF-1α, using digoxin and 2-methoxyestradiol, has been shown to ameliorate angiotensin II induced AAA [28]. Rapamycin, an inhibitor for mTOR signaling and HIF-1α accumulation [29], can attenuate TAA in Fbn1^C1041G/+, via suppressing ERK1/2 [30]. Rapamycin also ameliorates β-aminopropionitrile-induced TAA and pancreatic elastase-induced AAA via reducing MMP2 and MMP9 and mural macrophage infiltration [31]. Contrary, a smooth muscle cell specific Hif-1α knockout displayed detrimental effects on β-aminopropionitrile and angiotensin II-induced aortic TAA and AAA by reducing lysyl oxidase and tropoelastin mRNA [32]. These data suggest that elevated HIF-1α is closely associated with aneurysm development, but the underpinning mechanism is not clear. Moreover, whether and how hypoxia influences the inflammatory responses or MMP secretion in aortic smooth muscle cells during TAA pathogenesis is not well-defined.

miRNAs constitute a family of small non-coding RNAs which promote degradation and suppress translation of target mRNA [33]. miRNAs constitute a regulating network essential for aortic homeostasis. miR-29b is upregulated in the ascending aorta of Fbn1^C1041G/+ mice [34]. Inhibition of miR-29b ameliorates aneurysmal progression in both MFS and AAA mouse models by rescuing the expression of multiple extracellular matrix (ECM) genes, including COL3A1 [34, 35]. Lentiviral overexpression of miR-21 and miR-24 slowed AAA progression, by targeting the mRNA of phosphatase and tensin homolog protein and chitinase 3-like 1, respectively [36, 37]. These data show that some miRNAs are important players in aneurysm progression. Furthermore, our previous studies demonstrated that fibrillin-1–cell interaction regulates a subgroup of miRNAs in an “outside-in” manner, which in turn guides cell proliferation, focal adhesion, and TGF-β signaling [38, 39]. However, whether and how fibrillin-1-controlled miRNAs participate in TAA pathogenesis in MFS is not yet explored. In this study, we describe a fibrillin-1-controlled miRNA as a novel link integrating fibrillin-1–cell interaction and excessive HIF-1α with inflammatory responses and MMP12 expression during TAA pathogenesis.

Results

Baseline characterization of the ascending aortic aneurysms in Fbn1^mgR/mgR mice

To study the miRNA and mRNA expression patterns during TAA pathogenesis in MFS, we used Fbn1^mgR/mgR mice, a fibrillin-1 hypomorph MFS mouse model [15]. For baseline characterization, we first investigated whether male and female Fbn1^mgR/mgR mice developed aortic aneurysms similarly or differently. Gross views of freshly isolated thoracic aortae at 4 weeks and 10 weeks of age from both sexes were recorded (Fig. 1A–D). At 4 weeks of age, male Fbn1^mgR/mgR mice already displayed ascending aortic dilations (1.2-fold enlargement) compared to wild-type mice, but not the female mice. At 10 weeks of age, the aortic dilation in male Fbn1^mgR/mgR mice developed into aneurysms (2.1-fold enlargement). For female Fbn1^mgR/mgR mice at 10 weeks, the diameters were 1.5-fold larger than in wild-type mice. These results demonstrate that male Fbn1^mgR/mgR mice presented with earlier onset and faster progression of TAA than female mice. Therefore, we focused in this study on male mice. Figure 1E shows representative cross sections of ascending aortae of 10-week-old male wild-type and Fbn1^mgR/mgR mice. The figures present the autofluorescence originating primarily from elastic laminae, and fibrillin-1 staining of the tunica media. Consistent with the gross anatomical analyses, the lumen of the ascending aortae was 2.1-fold larger in Fbn1^mgR/mgR mice than in wild-type mice, and the aortic tunica media were 1.6-fold wider (Fig. 1F), showing the expected elastic lamina fragmentation (Fig. 1E, white arrowheads). Consistent with the hypomorphic Fbn1 gene expression, fibrillin-1 staining of the ascending aortae decreased at the aneurysmal site of Fbn1^mgR/mgR mice to 26% of the wild-type level, close to the reported 20–25% Fbn1 mRNA expression in skin and lung [15].

Fig. 1.

Baseline characterization of ascending aortic aneurysms in Fbn1^mgR/mgR mice. A and C Gross view of freshly isolated thoracic aortae from male (A) and female (C) wild-type (WT) and Fbn1^mgR/mgR mice at 4 and 10 weeks of age. Arrows point to the dilated region of the ascending aortae. B and D Quantification (diameters) of aneurysms in the ascending aortae as shown in A and C, respectively. Each data point represents the measurement from one mouse. Total numbers for male tissues: 4-week WT n = 4, 4-week Fbn1^mgR/mgR n = 4, 10-week WT n = 8, 10-week Fbn1^mgR/mgR n = 9. Total numbers for female tissues: 4-week WT n = 4, 4-week Fbn1^mgR/mgR n = 4, 10-week WT n = 4, Fbn1^mgR/mgR n = 4. E Representative images of elastic lamina auto-fluorescence (Elastin) and fibrillin-1 staining (FBN1) of ascending aortic cross sections from 10-week-old male wild-type and Fbn1^mgR/mgR mice. White arrowheads indicate events of fragmentated elastic laminae shown by the autofluorescence in green. Nuclear DAPI staining is shown in blue. F Quantification of the lumen area and media thickness determined from cross sections shown in (E) as described in detail in the Methods. Intensities of FBN1 staining were determined by ImageJ and normalized to the cell number (DAPI). The average mean intensity of FBN1 staining per cell in WT tissue was set to 1. Each data point represents the analysis of one mouse. Total sample numbers for lumen area and media thickness: WT n = 11, Fbn1^mgR/mgR n = 7. For FBN1 quantification, a total of WT n = 5, and Fbn1^mgR/mgR n = 4 were analyzed. Two sample t-test or Mann–Whitney U test was performed for normally distributed or non-normal distributed data, respectively. *p values less than 0.05

Time-dependent miRNA profiling of ascending aortae of Fbn1^mgR/mgR mice

Our previous studies have shown that the interaction of cells with fibrillin-1 regulates miRNAs [38, 39]. These data suggest that miRNAs could be differentially regulated in Fbn1^mgR/mgR aortae, which are characterized by deficient fibrillin-1 deposition in the media (see Fig. 1E, F). To explore whether and how miRNAs are differentially regulated during progression of aortic dilation, two timepoints were chosen to perform miRNA microarray analyses in male mice based on the initial characterization at the early stage of aortic dilation (4 weeks) and at the fully developed aneurysm stage (10 weeks). At 4 weeks, 3 miRNAs were downregulated and 14 upregulated more than twofold (Fig. 2A, C). At 10 weeks, the number of differentially regulated miRNAs (> twofold) increased to 129 with 5 down- and 124 upregulated miRNAs (Fig. 2B). Figure 2D shows the top 3 down- and top 20 up-regulated miRNAs in 10-week ascending aortic tissue. Bioinformatic prediction using mirPath v.3 revealed inflammatory responses (red) and ECM-related pathways (green) among the top hits regulated by the differentially expressed miRNAs at both early and late stages of aneurysm formation (Fig. 2E, F). miR-122 was the most downregulated miRNA at 10 weeks with 5.8-fold lower expression in Fbn1^mgR/mgR compared to the wild-type tissues (Fig. 2B). To include older aortic tissues, which is not possible for Fbn1^mgR/mgR mice because of lethality around 3–4 months, we additionally analyzed 70-week-old Fbn1^C1041G/+ mice. miR-122 was also significantly downregulated (~ 2.5-fold lower) in ascending aneurysmal aortae of these mice (Supplementary Fig. 1A). The data strongly suggested a role of miR-122 in TAA pathogenesis in these MFS mouse models.

Fig. 2.

Time-dependent miRNA profiling of ascending aortae of wild-type and Fbn1^mgR/mgR mice. A and B Volcano plots showing differentially expressed miRNAs in the ascending aorta of Fbn1^mgR/mgR versus wild-type (WT) mice at 4 and 10 weeks of age detected by Affymetrix GeneChip miRNA 4.0 Arrays. Green data points represent upregulated and red data points indicate downregulated miRNAs (> twofold, p < 0.05). 4 wild-type and 4 Fbn1^mgR/mgR mice were analyzed. miR-122, the most downregulated miRNA is indicated by a black circle. C Levels of all 17 differentially expressed miRNAs (> twofold) at 4 weeks of age from (A) were plotted. D Top 3 downregulated and top 20 upregulated miRNAs at 10 weeks of age from (B). Data in (C and D) are shown as fold change Fbn1^mgR/mgR versus WT. E and F Pathway enrichment prediction using all differentially expressed (> twofold, p < 0.05) miRNAs at 4- (E) and 10 weeks (F) of age, using the DIANA tool mirPath v.3 (http://snf-515788.vm.okeanos.grnet.gr) [68]. The inflammation-related pathways are indicated in red. ECM/cell adhesion relevant pathways are in green

mRNA profiling of ascending aortae of Fbn1^mgR/mgR mice

To determine differentially regulated mRNAs, we performed mRNA microarray analyses of aortic aneurysm tissue isolated from 10-week-old male wild-type and Fbn1^mgR/mgR mice (Fig. 3A). A total of 471 mRNAs were up-regulated (green) and 253 downregulated (red) more than twofold (p < 0.05) in aneurysmal versus wild-type tissues. Gene ontology analyses using all 724 differentially regulated mRNAs are shown in Fig. 3B, C. As was predicted from miRNA profiling, a prominent upregulation of inflammatory-related pathways was observed, including cytokine–cytokine receptor interaction as the top enriched pathway that included 26 genes (Fig. 3B; red labels). Multiple ECM relevant pathways were also upregulated in Fbn1^mgR/mgR aortae (Fig. 3B; green labels).

Fig. 3.

Global mRNA analyses and selected protein levels relevant to inflammation in the ascending aortae of 10-week-old Fbn1^mgR/mgR versus wild-type mice. A Volcano plot of differentially expressed mRNAs of ascending aortae of 10-week-old Fbn1^mgR/mgR (n = 4) versus wild-type (WT, n = 4) mice analyzed by Affymetrix Clariom D array (mouse). Green data points represent upregulated and red data points indicate downregulated mRNAs (> twofold, p < 0.05). B and C Gene Set Pathway Enrichment (GSEA; https://www.gsea-msigdb.org [67]) analyses showing the upregulated (B) and downregulated (C) pathways. All > twofold (p < 0.05) differentially regulated mRNA transcripts detected were used for the analysis. Red text indicates inflammatory relevant pathways, green text indicates ECM/cell adhesion-related pathways. D Heat maps of the 9 most upregulated mRNAs in the “Cytokine–cytokine receptor interaction” pathway shown in (B), and all altered matrix metalloproteinases in the aortae of 4 Fbn1^mgR/mgR mice at 10 weeks of age (right panels) and 4 WT shown in the left panels. E Western blotting and quantification of CCL2 and IL1β levels in 10-week-old WT and Fbn1^mgR/mgR aortae. Each data point represents the analysis of one mouse, a total of 4 WT and 5 Fbn1^mgR/mgR mice were analyzed. F Representative immunofluorescence images of MMP12 staining of ascending aorta sections from 10-week-old Fbn1^mgR/mgR and WT mice. Quantification of MMP12 intensity per area normalized to the signals of WT are shown on the right, with WT n = 5, Fbn1^mgR/mgR n = 10. Every data point represents one mouse. Error bars represent standard deviation. Two-sample t-test was used for statistical analyses in (E). Mann–Whitney test was used to test significance for the statistical analyses in (F). Error bars represent standard deviations. *p values less than 0.05

Heatmaps of the top upregulated mRNAs in the “Cytokine–cytokine receptor interaction” pathway from Fig. 3B and all the differentially regulated MMPs are shown in Fig. 3D. Ccl2 (4.1-fold), Il1β (2.9-fold) and Mmp12 (5.1-fold) were among the top upregulated mRNAs. Elevation of Ccl2 (3.9-fold), Il1β (8.0-fold) and Mmp12 (4.2-fold) was also observed in 70-week-old Fbn1^C1041G/+ ascending aortae (Supplementary Fig. 1B). Western blotting of freshly isolated 10-week-old Fbn1^mgR/mgR ascending aortae confirmed that CCL2 and IL-1β were upregulated at the protein level, 3.8 and 2.1-fold, respectively (Fig. 3E). Immunofluorescence staining confirmed the upregulation of MMP12 (1.7-fold) in aneurysmal tissues from Fbn1^mgR/mgR mice (Fig. 3F). As the inflammatory responses were upregulated in enriched pathway analyses of 10-week ascending aortic tissues of Fbn1^mgR/mgR mice, we investigated immune cell infiltration at that timepoint. Consistent with studies on human TAA patient tissues [14], macrophages (CD68) and T cells (CD3) were present in the aortic adventitia of the Fbn1^mgR/mgR ascending aorta, but not in tissue from wild-type mice (Supplementary Fig. 2, white arrow heads).

RNA-seq analyses of TAA tissues from human patients

To identify differentially regulated pathways in human TAA tissues, RNAseq analyses were performed to analyze 5 tissue samples obtained from TAA patients (Supplementary Table 1), compared to 3 non-aneurysmal aortic control samples from donors with no evidence of aortic disease (Fig. 4). Consistent with the data obtained from Fbn1^mgR/mgR mice, gene ontology analyses of the 4761 differentially (> twofold; p < 0.05) regulated mRNA transcripts showed that the “Cytokine–cytokine receptor interaction” pathway (Fig. 4A; red text) was the top hit that included 39 upregulated mRNAs. ECM-related pathways were also differentially regulated (Fig. 4A, B; green text). Comparative analyses of the differentially regulated mRNAs from Fbn1^mgR/mgR mice and from human TAA samples revealed 8 common upregulated and 5 common downregulated pathways (Fig. 4C, D). “Cytokine–cytokine receptor interaction” was the top commonly upregulated pathway in TAA tissues from Fbn1^mgR/mgR mice and human patients with the largest number of altered mRNAs (26 in Fbn1^mgR/mgR mice and 39 in human patient tissues) (Fig. 4E; red text). The green highlighted pathways in Fig. 4E show commonly upregulated ECM-related pathway. These comparative analyses confirmed the overlapping upregulation of inflammatory responses and ECM-related pathways in aneurysmal aortae of both Fbn1^mgR/mgR mice and human TAA patients. We next sought to elucidate whether miRNAs were involved in regulating these altered pathways.

Fig. 4.

RNA-seq analyses of human TAA samples and comparison with Fbn1^mgR/mgR. A and B RNA-seq analyses was performed on 5 human TAA samples and compared with 3 normal control samples as described in the Methods. Results of GSEA analyses using all > twofold (p < 0.05) differentially expressed mRNA transcripts are indicated for the upregulated (A, green bars) and the downregulated (B, red bars) KEGG pathways. Inflammatory relevant pathways labels are colored red. ECM-related pathways are indicated by green text. C and D Venn diagrams comparing common and different pathways based on GSEA analyses of Fbn1^mgR/mgR aortic samples (beige) and human TAA samples (blue). 8 pathways were commonly upregulated (C), and 5 pathways were commonly downregulated (D). E and F Overlapping pathways identified in (C and D) are shown with more details in (E and F), respectively. The number of altered genes in each pathway is shown for Fbn1^mgR/mgR-derived tissue (beige bars) and for human TAA-derived tissue (blue bars)

Comparative analysis of miRNAs and mRNAs levels in aortic aneurysm tissues

Involvement of miRNAs in TAA pathogenesis has been shown in regulating aortic wall apoptosis, and mRNA expression of elastin and collagen III in Fbn1^C1041G/+ mice [34]. However, little is known whether miRNAs participate in the regulation of the inflammatory response and other ECM relevant pathways during TAA pathogenesis. As stated above, bioinformatic analyses predicted the involvement of differentially regulated miRNAs in inflammatory responses and ECM relevant pathways in aneurysmal aortae of Fbn1^mgR/mgR mice (see Fig. 2C–F). To further understand which mRNAs were regulated by these miRNAs, we compared the expression levels between the differentially regulated miRNAs and mRNAs detected by microarrays of aortic tissues harvested from 10-week-old Fbn1^mgR/mgR mice. Since the primary function of miRNAs is to promote the degradation of mRNA targets [40], the levels of miRNAs and their mRNA targets typically inversely correlate. Thus, we compared the expression levels of miRNAs and their experimentally validated mRNA targets (based on the miRNA Targets Database TarBase, which contains experimentally validated miRNA–mRNA interactions [41]) and selected the inversely correlated pairs. Comparative analyses showed 89 miRNAs (differentially expressed > twofold) inversely correlated with 109 of their validated mRNA targets (> twofold). These differentially regulated miRNAs and mRNAs constitute 69% and 15% of the total differentially regulated miRNAs or mRNAs, respectively (Fig. 5A, green segments). Pathway enrichment analyses of the 109 mRNAs revealed that inflammatory-related pathways were the top hits (Fig. 5B; red text), including 5 differentially expressed mRNAs, Ccl2, Il1β, Cxcl13, Il6 and Bmpr1b. All of these, except Bmpr1b, were upregulated in the ascending aortae of Fbn1^mgR/mgR mice (Fig. 5C). Based on the TarBase library, miR-122, the most downregulated miRNA in Fbn1^mgR/mgR ascending aortae, targets all 4 upregulated inflammatory-related mRNAs. This strongly suggests the involvement of miR-122 in regulating inflammatory responses at the sites of dilated aortic walls.

Fig. 5.

Functional consequences of miR-122 downregulation. A Pie chart categorizing the differentially expressed 129 miRNAs and 724 mRNAs at 10 weeks (> twofold, p < 0.05) based on whether they are inversely correlated. Green sectors indicate that the expression of 89 miRNAs inversely correlated with 109 of their mRNA targets. B KEGG pathway analyses of the 109 mRNAs which inversely correlated with the miRNAs that targets them (green sector in (A)), using GSEA. Only upregulated pathways are shown. Inflammatory-related pathways are indicated by red text. C Heat map of the differentially regulated mRNAs in the “Cytokine–cytokine receptor interaction” pathway from (C). Data are shown for 4 wild-type (WT; left) and 4 Fbn1^mgR/mgR (right) mice. D In situ hybridization of miR-122 (white signal) in ascending aortic cross sections from 10-week male WT and Fbn1^mgR/mgR mice. Quantification of the miR-122 intensity is shown on the right. The signal intensity was normalized to the cell numbers. Each data point represents one animal, with WT n = 5 and Fbn1^mgR/mgR n = 13. Error bars represent standard deviations. The two-sample t-test was used to test significance. E qPCR analyses of the 4 upregulated mRNA targets of miR-122 (shown in C) and Mmp12 using total RNA from ex vivo aorta organ cultures treated with miR-122 inhibitor (miR122IH) compared to scrambled controls. Aortae from wild-type mice were used. One part of the ascending aorta was treated with miR-122 inhibitor to mimic miR-122 levels in the aneurysmal aorta, and the other part from the same ascending aorta was used for treatment with the scrambled control. Aortae from 8 to 9 WT type mice were used. Linked are the data points for the aorta from each mouse treated with either inhibitor or with the control. Based on the data distribution, paired sample t-test was performed for Ccl2, Wilcoxon signed rank tests were used for Il1β, Mmp12, Cxcl13 and Il6 to test significance. F Multiplexed ELISA analyses of 178 small secreted bioactive factors present in the conditioned medium of human aorta smooth muscle cells upon treatment with miR-122 inhibitor (miR122IH) or scrambled control for 48 h. The pie chart indicates the number of upregulated (7, blue), downregulated (1, beige), and unaltered (35, orange) factors upon miR-122 inhibition. The gray sector represents tested secreted factors below the detection level (135). G Bar charts indicate the relative protein levels of CCL2, IL-1β and MMP12. A total of 6 technical replicates normalized and pooled from two experiments (3 for each) were analyzed. For statistical analyses, the two-sample t-test was used. *p values less than 0.05

Functional analyses of miR-122 on inflammatory responses and matrix metalloproteinase expression

In situ hybridization of ascending aortae from 10-week-old wild-type male mice showed that miR-122 was primarily present in smooth muscle cells of the media with particular strong staining patterns adjacent of the elastic laminae (Fig. 5D). In ascending aortae of age and sex matched Fbn1^mgR/mgR mice, miR-122 was significantly downregulated to 49%, consistent with the data obtained by miRNA arrays. In situ hybridization without the probe or with a scrambled probe showed no signal in the aortae, which validated specificity of the miR-122 hybridization (Supplementary Fig. 3). Functional consequences of downregulated miR-122 on the mRNA expression levels of Ccl2, Il1β, Cxcl13, and Il6 were investigated with ex vivo aorta organ cultures from 10-week wild-type mice. Fresh aortae from the same animal were cut into 1 mm small pieces. To mimic the low miR-122 levels in aneurysmal tissue, one part of the ascending aorta was transfected with miR-122 inhibitor (miR122IH), and the other part was transfected with scrambled controls (Fig. 5E). The Ccl2 and Il1β mRNAs were significantly upregulated by 2.1- and 4.4-fold, respectively, upon miR-122 inhibition, but not the Cxcl13 and Il6 mRNA. Surprisingly, although Mmp12 mRNA is not predicted to be a target of miR-122, it was also upregulated (4.1-fold) upon miR-122 inhibition in ex vivo organ cultures (Fig. 5E). To validate these data, we conducted multiplexed ELISA experiments analyzing 178 soluble secreted factors including cytokines and MMPs from cultured human aortic smooth muscle cells after miR-122 inhibition (Fig. 5F, G and Supplementary Fig. 4). We identified 7 upregulated, 1 downregulated and 35 unaltered factors, whereas the remaining tested factors were below the detection limit (Fig. 5F). Consistent with the mRNA data, among the 7 upregulated proteins were CCL2 (1.3-fold), IL-1β (1.3-fold) and MMP12 (1.4-fold) (Fig. 5G). Other upregulated proinflammatory cytokines or receptors include IL-1α, granulocyte–macrophage colony-stimulating factor (GM-CSF), growth differentiation factor 15 (GDF-15), and tumor necrosis factor receptor 1 (TNFR1) (Supplementary Fig. 4). These results suggested that the downregulated miR-122 in TAA tissue of Fbn1^mgR/mgR mice promoted the inflammatory response by post-transcriptional upregulation of proinflammatory cytokines CCL2 and IL-1β, and by the elastic fiber degrading enzyme MMP12, which is also involved in inflammatory responses [42].

In addition, miR-122 was predicted to target mouse as well as human fibrillin-1 mRNA at 3 sites within the open reading frame (Supplementary Fig. 5A). The molecular interactions of miR-122 with the predicted binding sites were validated using dual luciferase assays, resulting in reduced firefly luciferase activities by 10–50% (Supplementary Fig. 5B, C). These data suggest that reduced miR-122 in TAA tissue may partially elevate the hypomorph fibrillin-1 mRNA in the Fbn1^mgR/mgR aortae. This is consistent with the observation that the fibrillin-1 mRNA was only reduced to 65% of normal levels in Fbn1^mgR/mgR aortae (Supplementary Fig. 5D), instead of reduction to the expected 20–25% reported by Pereira et al. for skin and lung [15]. However, this partial rescue of mRNA transcripts did not translate into increased microfibril levels, as the deposited fibrillin-1 protein was only 26% compared to wild type, as shown in Fig. 1E, F. Together, we interpret the data as a non-productive attempt of the aortic smooth muscle cells to rescue fibrillin-1 deficiency in the ECM. It is unclear why it remains unproductive, but crosslinked microfibrils possibly prevent further incorporation of fibrillin-1.

Deficient fibrillin-1 downregulates miR-122, possibly through c-Src signaling

Our previously published data have validated that a deficient fibrillin-1–fibroblast interaction regulates a subset of miRNAs [39]. Re-analysis of the microarray data from that study showed that miR-122 was downregulated by 20% after fibroblasts were seeded for 24 h onto rF1M-RGA, a central fibrillin-1 fragment (D⁹¹⁰–Q²⁰⁵⁴) not able to promote fibrillin-1–integrin interaction, as compared to the wild-type fibrillin-1 fragment rF1M-WT containing the RGD cell-binding sequence (Fig. 6A). Using an identical experimental setup with human aortic smooth muscle cells, we identified a larger decrease (71%) of miR-122 when the cells were seeded on rF1M-RGA for 24 h (Fig. 6B), as compared to rF1M-WT. Since the RGA mutation does not represent a known MFS mutation, and aortic smooth muscle cells from these MFS patients were not available for this study, we extended the experimental setup using cell-derived matrices produced by skin fibroblasts obtained from 7 MFS patients with defined mutations in fibrillin-1 (Supplementary Table 2). After production of the cell-derived matrices by the MFS fibroblasts or by fibroblasts from healthy controls for 7 days, the cells were removed from the matrices, followed by re-seeding of skin fibroblasts from healthy control subjects. The cell-derived matrices from the MFS fibroblast showed deficient deposition of fibrillin-1 in the matrix (Supplementary Fig. 6). qPCR analyses of the fibroblasts after 24 h revealed that miR-122 was downregulated by 61%, demonstrating that matrices containing a fibrillin-1 MFS mutation cannot maintain miR-122 levels, at least in fibroblasts (Fig. 6C). Similar experiments using smooth muscle cells failed, because these cells did not adhere to the MFS fibroblast-derived matrices. We next investigated potential kinases involved in regulating miR-122 when smooth muscle cells were seeded on culture dishes. Inhibitors of focal adhesion kinase (FAK) and c-Src kinase, which are known to be activated by integrin–ECM interactions [43], were applied to human aortic smooth cells for 48 h. Inhibition of c-Src, but not FAK, suppressed miR-122 levels in smooth muscle cells seeded on uncoated culture plates (Fig. 6D). Together, these data demonstrate that mutations in fibrillin-1 lower the levels of miR-122. The data also suggest that c-Src kinase is needed to maintain miR-122 levels under normal culture conditions.

Fig. 6.

Upstream regulators of miR-122. A Re-analysis of previous microarray data, comparing miR-122 levels of fibroblasts 24 h after seeding on rF1M-WT or rF1M-RGA [39]. 4 technical replicates for each condition were used. B qPCR analyses of miR-122 extracted from smooth muscle cells 24 h after seeding on rF1M-WT and rF1M-RGA. A total of 6 technical replicates normalized and pooled from two experiments (3 for each) were analyzed. C miR-122 qPCR analyses of normal skin fibroblasts seeded for 24 h either on cell-derived matrices from MFS fibroblasts derived from 7 patients (Supplementary Table 2), or on the matrices produced by fibroblasts derived from 3 healthy controls. D miR-122 qPCR analyses of smooth muscle cells upon treatment with inhibitors for FAK and c-Src kinase, with DMSO (1:1,000 v/v) as control. Each data point represents one technical replicate, from 2 experiments. E Relative Hif-1α mRNA levels determined by microarrays of 10-week-old male mice, comparing Fbn1^mgR/mgR (n = 4) to wild-type (WT, n = 4). F HIF-1α immunostaining of ascending aortic cross sections of 10-week-old male Fbn1^mgR/mgR and WT mice. The lumen side is positioned on the left (labelled L). Quantifications of HIF-1α staining are shown on the right. Each data point represents analysis of one animal comparing Fbn1^mgR/mgR (n = 6) to wild-type (WT, n = 4) mice. G miR-122 qPCR of human smooth muscle cells cultivated for 48 h under normoxic (Control) or 5% O₂ hypoxic conditions. A total of 6 technical replicates normalized and pooled from two experiments (3 for each) were analyzed. H miR-122 qPCR of aorta organ cultures from 9 wild-type mice cultivated for 72 h under normoxic or hypoxic conditions. Linked are the data points for the aorta from each mouse under either normoxia (control) or hypoxia. I miR-122 levels of human smooth muscle cells treated under hypoxic conditions (5% O₂) for 48 h with 1 µM digoxin, 500 nM 2-methoxyestradiol (2ME), or with the DMSO control (1:1000 v/v). For the digoxin experiments, one representative experiment with n = 4 for each condition is shown. For the 2ME experiments, one representative experiment with n = 3 is shown. Experiments were repeated with similar results. Two-sample t-test was used for statistical analyses in (A–G and I). Error bars represent standard deviations. Paired-sample t-test was used for statistical analysis in (H). *p values less than 0.05

Hypoxic conditions suppress miR-122 and fibrillin-1 network formation

Previous studies identified oxidative stress [23], and elevated HIF-1α protein levels at sites of human acute thoracic aneurysm dissections [26, 44]. Consistent with this finding, HIF-1α mRNA and protein levels were upregulated by fibrillin-1 deficiency in the ascending aortae of Fbn1^mgR/mgR mice (Fig. 6E, F). The mitochondrial respiratory chain relies on oxygen for energy production via oxidative phosphorylation [45]. Under hypoxic conditions, cells switch from oxidative phosphorylation to glycolysis as a means of ATP production [46]. Although HIF-1α mRNA was not upregulated based on the human TAA RNAseq data, it is noteworthy that “Oxidative phosphorylation” was one of the most downregulated pathways in human TAA with 34 downregulated genes (Fig. 4B, underlined). The upregulated HIF-1α in the Fbn1^mgR/mgR aneurysmal aortae and the predicted impaired oxidative phosphorylation in human TAA tissues indicate that the TAA tissues experienced hypoxic stress. To understand whether hypoxic conditions and activated HIF-1α affect miR-122, we investigated its expression levels in human aortic smooth muscle cell culture by reducing the atmospheric oxygen to 5% for 48 h (Fig. 6G). qPCR analyses showed that the miR-122 levels in a hypoxic atmosphere were suppressed by 85% as compared to normoxic conditions. qPCR analyses of ex vivo aorta cultures from 10-week-old wild-type mice showed 33% reduction of miR-122 levels upon 72 h of hypoxia treatment (Fig. 6H). As there is no specific inhibitor available for the HIF-1α protein, two pharmacologically available non-specific HIF-1α inhibitors, digoxin and 2-methoxyestradiol (2ME), were used to treat human smooth muscle cell cultures under hypoxic conditions in an attempt to rescue downregulated miR-122 (Fig. 6I). 48 h digoxin or 2ME treatment indeed rescued the effects by hypoxia resulting in upregulated miR-122 levels by 2.6-fold and 1.3-fold, respectively. These results demonstrate that HIF-1α negatively regulates miR-122 levels in smooth muscle cells promoting TAA.

Digoxin normalizes miR-122 levels and downstream targets in vivo which rescues aortic dilation

Since the HIF-1α inhibitor digoxin elevated miR-122 levels in smooth muscle cell culture experiments, we tested whether it is also efficacious in Fbn1^mgR/mgR mice. We administered digoxin (2 mg/kg) or the dimethyl sulphoxide (DMSO) control intraperitoneally daily from 5 to 10 weeks (Fig. 7A). HIF-1α levels in the tunica media of dilated ascending aortae were markedly decreased by 30% in digoxin-treated mice compared to the DMSO controls (Fig. 7B). In situ hybridization showed 1.7-fold elevated miR-122 staining and qPCR analysis demonstrated sixfold upregulation of miR-122 in digoxin-treated Fbn1^mgR/mgR aortae versus DMSO-treated controls (Fig. 7C, Supplementary Fig. 7). mRNA levels of the downstream miR-122 targets, Ccl2, Il1β and Mmp12, were then determined by qPCR (Fig. 7D). Ccl2 and Mmp12 mRNAs were downregulated by 93% and 70%, respectively, whereas the level of Il1β did not change. Immunofluorescence staining of CCL2 and MMP12 in the tunica media of ascending aortae documented downregulation at the protein level by 21% and 34%, respectively (Fig. 7E, F). As MMP12 is an elastase, elastic lamina fragmentation was quantified (Fig. 7G). Histological analysis of ascending aorta cross sections after 35 days of treatment showed that digoxin treatment resulted in less elastic–fiber fragmentation, with a 24% reduction of the breaks in the elastic laminae. Determining the size of the aneurysms at the experimental endpoint at 10 weeks showed that the diameter of the ascending aortae from Fbn1^mgR/mgR mice were significantly reduced from 1.7 to 1.3 mm upon digoxin treatment, similar to the size of digoxin treated wild-type mice (Fig. 7H).

Fig. 7.

Digoxin ameliorates TAA development in Fbn1^mgR/mgR mice. A Schematic of the experimental design: 5-week-old mice were treated daily with either 2 mg/kg digoxin or with 2% v/v DMSO (buffer control) in 100 μL of PBS via intraperitoneal injection for 5 weeks (red line), and ascending dilated aortae were harvested upon euthanasia at 10 weeks of age. B Representative immunofluorescence staining and signal quantification of the ascending aortae for HIF-1α (DMSO n = 6; digoxin n = 11). C In situ hybridization of miR-122 (DMSO n = 8; digoxin n = 7). D qPCR analyses of Ccl2, Il1β and Mmp12 mRNA levels in the ascending aortae of treated mice (DMSO n = 13; digoxin n = 9). E Representative immunofluorescence staining and quantification of CCL2 (DMSO n = 10; digoxin n = 10), and F MMP12 (DMSO n = 7; digoxin n = 10) in ascending aortae of the treated mice. G Quantification of breaks in the elastic laminae (red triangles) from tropoelastin immunostaining images, were quantified and normalized to the area of the tunica media. 3–5 images were taken for each mouse, and each data point represents the average of one mouse (Fbn1^mgR/mgR DMSO n = 7; Fbn1^mgR/mgR digoxin n = 9; WT DMSO n = 3; WT digoxin n = 5). H Gross view of freshly isolated thoracic aortae from male Fbn1^mgR/mgR and WT mice. Quantification of the dilated ascending aortae is shown on the right. Each data point represents one mouse (WT DMSO n = 5; WT digoxin n = 8; Fbn1^mgR/mgR DMSO n = 14; Fbn1^mgR/mgR digoxin n = 10). For all quantifications, two sample t-test or Mann–Whitney U test was performed for normal or non-normal distributed data, respectively. *p values less than 0.05

Discussion

This study demonstrates that reduced miR-122 in the aortic wall of Fbn1^mgR/mgR mice represents an important factor connecting fibrillin-1 deficiency and hypoxia in the cellular microenvironment with elevated inflammatory responses and degradation of elastic laminae. This pathomechanism is active during progression of TAA and involves increased CCL2 and MMP12 downstream of miR-122 and disrupted upstream integrin signaling and HIF1α upregulation. Digoxin treatment of Fbn1^mgR/mgR mice elevated miR-122 and consequently reduced CCL2 and MMP12, and in turn counteracted elastic laminae fragmentation and ameliorated aortic dilation.

TAA was reported to occur more often in male compared to female individuals, including MFS patients [47]. Consistent with these reports, the data presented here show that male Fbn1^mgR/mgR mice developed aortic dilation earlier than female mice. Cook et al. demonstrated that TGF-β neutralizing treatment of TAA in MFS mice showed a time-dependent efficacy, as blocking TGF-β signaling at an early stage of aortic dilation worsened aortic outcomes in Fbn1^mgR/mgR mice, whereas treatment at later stages attenuated TAA formation [48]. This finding emphasized the importance of analyzing dysregulated miRNA pathways at early and late stages of aortic aneurysm progression. Ascending aortic dilation (1.2-fold) occurred as early as 4 weeks of age in male Fbn1^mgR/mgR mice. Already at this stage, the miRNA profiles predicted inflammatory and ECM-related pathways to be differentially regulated, which was further consolidated at the late stage of 10 weeks. Whether inflammatory responses at the early stage are indeed altered at the mRNA and protein levels, and whether inflammatory responses at this timepoint are beneficial or detrimental for aorta homeostasis still requires further analyses. Nevertheless, the results suggested the involvement of inflammatory and ECM-related pathways in TAA pathogenesis starting from early stages of aortic dilation.

Elevated inflammatory responses persisted to the late dilation stage in the ascending aortae of Fbn1^mgR/mgR mice. Upregulated inflammatory responses were also observed in human TAA patients, despite some species-specific differences in the altered gene profiles. However, only relatively few inflammatory cells (macrophages and T-cells) were observed in the adventitia of the aneurysmal aortae in Fbn1^mgR/mgR mice. This is consistent with previous reports on human TAA patients [14]. Ccl2 and Il1β mRNAs were elevated in the aneurysmal aortae of both Fbn1^mgR/mgR and Fbn1^C1041G/+ mice. CCL2 and IL-1β protein upregulation was validated in the aortae of 10-week-old Fbn1^mgR/mgR mice. As these cytokines typically act in a paracrine and autocrine fashion [49], the paucity of inflammatory cells in the tunica media suggests that the primary effector cells are the aortic smooth muscle cells. Previous reports have shown that elevated CCL2 and IL-1β lead to elevated expression of multiple pro-inflammatory genes and MMPs in aortic smooth muscle cells, as well loss of smooth muscle cell contractility. For example, IL-1β can induce the expression of CCL2 in aortic smooth muscle cells, and can promote secretion of multiple MMPs from smooth muscle cells, including MMP 1, 2, 3 and 9 [19]. In addition, CCL2 and, IL-1β can repress the expression of smooth muscle cell contractile markers, including αSMA, smooth muscle myosin heavy chain, calponin-1, and SM22 [19], which correlates with the downregulation of “Calcium signaling pathways” in the TAA tissues of both Fbn1^mgR/mgR mice and human patients observed in our study. Literature data and the results presented here strongly suggest that upregulation of CCL2 and IL-1β in the dilated aortic wall of MFS mice promotes inflammatory responses and aortic wall destruction, associated with suppressed smooth muscle cell contractility.

A critical hallmark of TAA progression is the fragmented elastic laminae in the tunica media, correlating with elevated MMPs, including MMP2, 9 and 12. Among those, MMP2 and MMP9 has been widely studied in MFS mice. Protein levels of active MMP2 and MMP9 were upregulated in Fbn1^mgR/mgR mice at 6 weeks of age, and doxycycline inhibition delayed TAA progression [21, 50]. However, Mmp2 and Mmp9 mRNA levels were not altered in our microarray data obtained from 10-week-old mice. This apparent discrepancy may originate from the different times the mice were analyzed between the studies. It is possible that Mmp2 and Mmp9 mRNA is indeed upregulated at earlier timepoints but is downregulated to baseline levels at 10 weeks. Alternatively, upregulation of MMP2 and MMP9 on the protein level may depend on an mRNA-independent mechanism, possibly reduced turnover of these enzymes. MMP12 is elevated in the aortae of MFS patients and Fbn1^mgR/mgR mice [51], and in the serum from type A acute aortic dissection patients without known genetic syndromes [52], which strongly suggests an important role of MMP12 in TAA progression. Consistent with these data, we observed elevated MMP12 in the TAA tissues of Fbn1^mgR/mgR mice, associated with decreased miR-122 levels. Although known as macrophage metalloelastase, MMP12 is also expressed in smooth muscle cells [53], which is consistent with our human smooth muscle culture data and aortic tissue immunofluorescence staining of aortae from Fbn1^mgR/mgR mice. Inhibition of miR-122 elevated MMP12 both in ex vivo aorta cultures and in human smooth muscle cell cultures. As MMP12 has a higher affinity to elastic fibers, as compared to other MMPs, including MMP9 [54], it is very likely that the enhanced elastic fiber degradation in the aortae of Fbn1^mgR/mgR mice is at least in part mediated by MMP12.

Consistent with the bioinformatic prediction in our study suggesting that miR-122 is an important miRNA regulating inflammatory pathways in TAA tissues, the global knockout of miR-122 led to increased inflammatory responses in the liver, with upregulated CCL2, IL-6 and TNF-α [55]. Our study demonstrates by in situ hybridization that miR-122 is present in the aortic wall, primarily localized to smooth muscle cells adjacent to elastic laminae in the tunica media. Importantly, our functional studies showed that inhibition of miR-122 led to increased CCL2, IL-1β and MMP12 using ex vivo aorta cultures and human smooth muscle cell cultures. The Ccl2 and Il1β mRNAs are confirmed as direct targets of miR-122 [56], whereas the Mmp12 mRNA is not a predicted miR-122 target. However, the in vitro, ex vivo and in vivo functional analyses clearly demonstrated that MMP12 is negatively regulated by miR-122 at both mRNA and protein levels. There are two possible explanations: (i) Mmp12 mRNA is indeed a direct target of miR-122 but is not predicted by current bioinformatic tools. The algorithms predict miRNA–mRNA interaction efficacies based on the primary linear miRNA and mRNA sequences [57]. It is possible, however, that target binding site(s) are constituted by non-contiguous sequences in the 3’ untranslated region of the Mmp12 mRNA, which is not predictable by the current tools, and difficult to address experimentally. (ii) Another possibility is that miR-122 does not directly target Mmp12 mRNA, but indirectly through other direct targets, such as CCL2 and IL-1β. Although there is no data available yet for MMP12, CCL2 and IL-1β have both been demonstrated to elevate multiple other MMPs in smooth muscle cells. For instance, CCL2 increased MMP9 in smooth muscle cells [58], and IL-1β can induce secretion of MMP1, 2, 3 and 9 in the smooth muscle cell culture medium [59]. Together, our data clearly revealed miR-122 as a regulatory miRNA of the inflammatory response and the integrity of elastic laminae in TAA tissues of Fbn1^mgR/mgR mice.

To understand the mechanism how miR-122 was downregulated in TAA tissues of Fbn1^mgR/mgR mice, we explored upstream effectors [13]. As a fibrillin-1 hypomorph mouse model, Fbn1^mgR/mgR mice are characterized by loss of connections between smooth muscle cells and elastic laminae in the tunica media mediated by fibrillin-1 microfibrils [22]. Fibrillin-1–cell interaction can regulate miRNA levels in fibroblasts [38, 39], and here we show that for aortic smooth muscle cells. Deficiency of fibrillin-1–cell interactions downregulated miR-122 in both fibroblasts and smooth muscle cells, involving c-Src kinase activity. Integrins α5β1 and αvβ3 are the primary transmembrane receptors for fibrillin-1 on mesenchymal cells [60, 61], which both activate c-Src upon ligand binding. Activation of c-Src by β1 integrins always involves FAK, whereas β3 integrins can also activate c-Src kinase directly, independent of FAK [43]. The data presented here showing that the miR-122 level was maintained in an FAK-independent manner in smooth muscle cells thus suggest the involvement of β3 integrin in regulating miR-122. Reduced levels of either normal fibrillin-1 in the Fbn1^mgR/mgR aortic tissue or in MFS-derived matrices should ligate less β3 integrins, leading to decreased miR-122 and the increased downstream inflammatory response and elastin fragmentation.

Deficiency of fibrillin-1 in the matrix can also induce Hif-1α expression in vascular smooth muscle cells [44]. The microarray and immunofluorescence data document increased HIF-1α in the aneurysmal aortae of Fbn1^mgR/mgR mice, which indicates a hypoxic environment in the tunica media. Human TAA tissue experience oxidative stress with elevated HIF-1α protein levels [26], which is supported by our RNAseq data showing suppressed oxidative phosphorylation in the human TAA tissues. Our study identified elevated HIF-1α as an upstream regulator of miR-122 in TAA tissues. Consistent with our study in smooth muscle cells and ex vivo aorta cultures, hypoxic conditions downregulated miR-122 expression in squamous cell carcinoma [62], and reoxygenation after hypoxia treatment of myocardial cells upregulated miR-122 [63]. As there is no specific pharmacological inhibitor available for HIF-1α, we used the well-established and widely accepted HIF-1α inhibitors digoxin and 2-methoxyestradiol (2ME) in this study, which do not directly target HIF-1α [28]. Consistent with the above mentioned reoxygenation experiments, both HIF-1α inhibitors could rescue downregulated miR-122 in smooth muscle cells cultivated under hypoxic conditions. Treating Fbn1^mgR/mgR mice with digoxin restored miR-122 levels in TAA lesions. Digoxin exerts anti-inflammatory and protective roles in AAA formation. For example, Wei et al. demonstrated that digoxin attenuated AAA formation in angiotensin II and pancreatic elastase perfused mouse models by inhibiting the presence of T helper cells and IL-17a signaling [64]. In addition, digoxin as well as 2ME reduced MMP2 and MMP9 levels in an angiotensin II induced hyperlipidemic AAA mouse model [28]. However, TAA and AAA represent different pathophysiologies of the aorta. Although both present with elastic laminae fragmentation, disrupted ECM and altered smooth muscle cell contractility in the tunica media, TAA tissue is characterized by less inflammatory cell infiltrations in the tunica media [14, 15] (see Supplementary Fig. 2), as compared to the large presence of inflammatory cells in AAA tissue [65]. One of the major outcomes of digoxin treatment in AAA models is the reduction of inflammatory cell infiltration, whereas in TAA tissues from Fbn1^mgR/mgR mice in our study, the primary effector cells are the tunica media smooth muscle cells. Our data demonstrated that digoxin restored miR-122 levels and decreased the amount of CCL2 and MMP12 in smooth muscle cells in vitro and in vivo. Thus, this study established miR-122 as a novel connector of hypoxia with inflammatory responses and MMP12 secretion by smooth muscle cells.

It is also worth noting that miR-122 was not downregulated at 4 weeks of age when mild aneurysmal dilation was already observed in male Fbn1^mgR/mgR mice, but it was significantly downregulated at 10 weeks of age at the full dilatation stage. This indicates that downregulated miR-122 cannot be the initiator of aneurysm formation but rather plays a role in disease progression. It is likely that both upstream regulators, increased HIF-1α and reduced fibrillin-1–cell interactions, do only play a role during disease progression. For example, increasing aggrecan aggregation in the aortic wall could progressively restrict oxygen diffusion leading to hypoxic conditions [66]. It is also possible that the fibrillin-1–cell interactions may be less affected in early stages, because increasing degradation of microfibrils and fibrillin-1 during aortic aneurysm progression may gradually affect this interaction.

In conclusion, the following working model for TAA progression in MFS emerges from this study (Fig. 8). Reduced or deficient fibrillin-1 in the cellular microenvironment leads to (1) compromised microfibrils interacting with smooth muscle cells, and (2) hypoxic conditions in the tunica media. Deficient fibrillin-1–cell interaction reduces miR-122 levels likely via decreased c-Src activity. Hypoxia-induced elevated HIF-1α also decreases miR-122 levels. Consequently, CCL2, IL-1β and MMP12 are upregulated on the mRNA and protein levels, because they are miR-122 targets. This leads to increased elastic lamina fragmentation and TAA progression. Treatment with digoxin inhibits HIF-1α and rescues miR-122 levels in the Fbn1^mgR/mgR aortae, resulting in reduction of elastic lamina fragmentation and TAA progression by suppressing CCL2 and MMP12.

Fig. 8.

Working model illustrating the central role of miR-122 in smooth muscle cells during TAA progression. Reduced or deficient fibrillin-1 leads via two functional arms to reduced miR-122 levels in the aortic aneurysmal wall. This leads via increased inflammatory cytokines and MMP12 to elastic lamina fragmentation and TAA progression. Digoxin treatment interferes at the level of HIF-1α and ameliorates TAA progression. Details of this model are explained in the last paragraph of the discussion

Materials and methods

Mice

Heterozygous mutant Fbn1^mgR/+ mice on a C57BL/6J genetic background were bred to generate wild-type and Fbn1^mgR/mgR mice for experiments. Fbn1^mgR/mgR mice represent a severe MFS model characterized by reduced normal fibrillin-1 levels of about 20–25% compared to wild-type mice [15]. Fbn1^C1041G/+ and wild-type mice on a C57BL/6J genetic background were purchased from the Jackson Laboratory and bred to generate mice with both genotypes. Fbn1^C1041G/+ is a milder MFS model expressing fibrillin-1 with a missense mutation [16].

Digoxin injection

Mice received digoxin (2 mg/kg/day, Sigma, Cat# D6003) via intraperitoneal injection starting from 5 weeks of age. Digoxin was dissolved at 250 mg/mL in DMSO, then diluted 1:50 in phosphate-buffered saline (PBS) to 12.5 mg/mL for injections. Equal volume of DMSO (1:50, v/v) in PBS was injected as buffer control. Treated mice were euthanized at 10 weeks of age. In total, 11 Fbn1^mgR/mgR mice were injected with digoxin and 14 were injected with DMSO as control.

RNA extraction and reverse transcription PCR

For RNA extraction from cells, 200,000 human smooth muscle cells grown in 12-well plates (Sarstedt) were lysed in 100 µL Qiazol (Qiagen). For RNA extraction from tissues, the ascending aortae were sonicated 3 × 5 s in 200 µL Qiazol on ice. Extraction of total RNA including both miRNAs and mRNAs was performed using the miRNeasy Mini Kit (Qiagen, Cat# 217004), according to the manufacturer’s instructions. For reverse transcription of RNA into cDNA, 0.1–1 µg of total RNA was used. miScript II RT kit (Qiagen) was employed to reverse transcribe miRNAs, according to the manufacturer’s instructions. Super Script III First-Strand Synthesis System for RT-PCR (Life Technologies) was used to reverse transcribe mRNAs as detailed in the supplier’s manual.

Microarray analysis of mouse ascending aorta

Microarray analysis was performed for miRNAs and mRNAs using the GeneChip miRNA 4.0 Array and the Clariom D Array, mouse from Affymetrix, respectively. RNA extraction, RNA quality control, cDNA preparation and labeling, and the arrays were conducted as standard procedures at the Genome Quebec Innovation Centre at McGill University. 4 ascending aorta samples in each group from 4 and 10-week-old wild-type and Fbn1^mgR/mgR mice were prepared as described above, and snap-frozen in liquid nitrogen to preserve the integrity of RNA. Data processing and statistical analyses were performed by the Transcriptome Analysis Console software (Affymetrix). Normalized intensities between the aortae from wild-type and Fbn1^mgR/mgR mice were compared using one-way ANOVA, followed by multi-testing correction using the Benjamini–Hochberg false discovery rate (FDR)-controlling procedure for all expressed genes. All p values indicated refer to FDR-adjusted p values.

Human sample collections and RNA isolation

TAA specimen from 5 male patients (age range of 29–63 years, Supplementary Table 1) were collected upon open aortic repair procedures at the NYU Langone Medical Center, New York, NY, USA. None of the patients had genetic collagen disorders or were diagnosed with MFS. Non-aneurysmal aortic control tissues were collected from donors with no evidence of aortic disease at autopsies provided by LiveOnNY organization, New York, NY, USA. Patient information was collected by a questionnaire or from the patient’s file at the hospital (Supplementary Table 1). Tissue samples were oriented, formalin fixed, paraffin embedded and sectioned or kept frozen at − 80 °C for RNA isolation.

RNA sequencing

Total RNA extractions of the human thoracic aortae specimen were performed using the RNeasy Fibrous Tissue Mini Kit according to the manufacturer’s instructions (Qiagen, Cat# 74704). Isolated RNA samples were processed using the Clontech Low Input Kit according to manufacturer’s instructions to prepare RNA-Seq libraries, followed by purification using AMPure beads (Agilent Technologies). The quality of RNA samples was verified by a Bioanalyzer (Agilent Technologies, Cat# G2939BA). The samples were analyzed on a HiSeq 2500 System (Illumina) as paired-end reads with 50 nucleotides in length. The reads were mapped against the hg19 human reference genome, using the Tophat 2.0.9. HTSeq 0.6.1 phyton framework and hg19 GTF gene annotation (UCSC database) to process BAM alignment files. The Bioconductor package DESeq2 (3.2) was used to identify differentially expressed genes. The Benjamin and Hochberg method was performed to control the false discovery rate. Transcripts that had adjusted p < 0.05 were considered to be differentially expressed.

Pathway enrichment analyses

To discover the network of regulators and the pathways associated with transcriptomic data, significantly dysregulated transcripts from mRNA microarray and RNAseq experiments (with fold change > 2, p < 0.05) were analyzed using the Gene Set Enrichment Analyses (GSEA) software [67]. To identify the miRNA-regulated pathways, DIANA mirPath v.3.0 was used to analyze the dysregulated miRNAs (> twofold, p < 0.05), DIANA TOOLS-mirPath v.3 (http://snf-515788.vm.okeanos.grnet.gr) [68]. Predictions of differentially regulated pathways with p < 0.05 are listed.

Ex vivo aorta cultures

Mice were first perfused with PBS, then thoracic aortae were dissected and washed with cold PBS two times and cut into ~ 1 mm long segments that were placed into serum-free DMEM/F12 (1:1) culture medium (Gibco), supplemented with 100 µg/mL penicillin, 100 µg/mL streptomycin, and 2 mM glutamine (PSG) at 37 °C in a 5% CO₂ atmosphere.

Cell culture

Primary human smooth muscle cells and primary skin fibroblasts from healthy human and from individuals with MFS were used in the experiments. The primary human smooth muscle cells were purchased from ScienCell Research Laboratories (Cat# 6110). Human healthy skin fibroblasts were isolated from the foreskin of healthy boys (2–5 years of age) following a standard circumcision procedure approved by the Montreal Children’s Hospital Research Ethics Board (PED-06-054), and skin fibroblasts isolated from the abdomen of healthy adults were used as healthy controls. MFS skin fibroblasts with seven different mutations were purchased from the Coriell Institute (Camden, NJ, USA) (Supplementary Table 2). For all experiments, primary cells were used between passages 4 and 10. Human smooth muscle cells were cultured in 231 culture medium with growth supplement SMGS (Gibco) and 10% fetal bovine serum (FBS). Primary skin fibroblasts were cultured in DMEM medium, supplemented with 10% FBS. All media were additionally supplemented with PSG. Cells were cultured at 37 °C in a 5% CO₂ atmosphere. To grow cells under hypoxic conditions, a 5% oxygen atmosphere generated in a NuAire incubator (NU-4950).

For analyses of skin fibroblasts and smooth muscle cells seeded on recombinant rF1M-WT and rF1M-RGA, the proteins were purified as previously described [39]. Cell culture plates or chamber slides were first pre-coated with 100 µg/mL poly-d-lysine for 2 h at room temperature, followed by coating of the recombinant proteins at a concentration of 25 µg/mL in Tris-buffered saline containing 2 mM Ca²⁺ overnight at 4 °C. This coating concentration was previously optimized for promoting cell adhesion [39]. We showed previously that poly-d-lysine does not affect integrin-triggered signaling [39]. HIF-1α inhibitors were used in concentrations recommended by the manufacturers to treat human smooth muscle cells under hypoxic conditions directly after cell seeding (1 μM digoxin, Sigma, Cat# D6003; 500 nM 2-methoxyestradiol, Selleckchem, Cat# S1233).

Generation of fibroblast-derived matrices

Primary fibroblasts used for generating cell-derived matrices included normal foreskin fibroblasts, adult abdominal skin, and skin fibroblasts from adult MFS patients (Coriell Institute, Supplementary Table 2). Fibroblasts were seeded on 6-well plates at a density of 500,000 cells/well. DMEM with PSG and 10% FBS medium were changed every other day for 7 d. After rinsing twice with PBS, the cell layers were lysed for 2–3 min with 1 mL prewarmed (37 °C) cell-extraction buffer (0.5% v/v Triton X-100; 20 mM NH₄OH in PBS) until no intact fibroblast could be observed under the microscope. The fibroblast-derived matrices were washed twice with PBS and kept in DMEM medium with PSG for further experiments.

miRNA transfection of cells and ex vivo aorta cultures

The inhibitor of miR-122 was commercially obtained from Qiagen (Anti-hsa-miR-122, Cat# MIMAT0000421). The inhibitor was prepared with HiPerFect transfection reagent (Qiagen, Cat# 301704) in serum-free DMEM medium. The solution was applied directly to the cells seeded on 6-well plates, or to the aorta cultures on 48-well plates, at a final concentration of 50 nM.

Quantification of miRNA and mRNA

For quantification of miRNAs, real-time qPCR was performed using the miScript SYBR Green PCR Kit (Qiagen, Cat# 218073), according to the manufacturer’s instructions. miScript Primer Assays were obtained from Qiagen for human miR-122 (Cat# MS0001526). RNU6 (Cat# MS00033740) and SNORD (Cat# MS00033705) were used as references. For quantification of mRNA, SYBR Green Select Master Mix (Applied Biosystems, Cat# 4472908) with 200 nM forward and reverse primers were used. Expression levels of target genes were normalized to GAPDH. The real-time cycler (Applied Biosystems, Step-One Real-Time qPCR system) was used to obtain amplification data. The delta–delta Ct method was used to compare different groups.

Dual luciferase assay

Detailed plasmid construction strategies were previously described [38]. Briefly, predicted miR-122 binding site sequences in both mouse (ENSMUST00000028633.12) and human (NM_000138.4) fibrillin-1 mRNAs were obtained using microT-CDS (http://diana.imis.athena-innovation.gr/DianaTools/index.php?r=microT_CDS/index) [69]. Two chemically synthesized single-stranded DNA sequences (Integrated DNA Technologies) containing the predicted sequences were annealed in Oligo Annealing Buffer (Promega, Cat# C838A). Each double stranded DNA sequence containing one predicted site was inserted into the pmirGLO vector (Promega, Cat# E133A), using T4 DNA ligase (Invitrogen, Cat# LS15224017). Each of the plasmids (2.5 ng/μL) were co-transfected using lipofectamine 2000 (Invitrogen, Cat# 11668–019) with either miR-122 mimic or a scrambled control (100 nM). Transfection was performed with 10,000 HEK293 cells per well in white 96-well plates (Greiner Bio-One Cat# 655073) in OPTI-MEM I Reduced Serum Medium (Gibco, Cat# 31985–062). The activities of firefly luciferase and Renilla luciferase were activated with the Dual-Luciferase Reporter Assay System (Promega, Cat# E2920) 48 h after transfection. Luminescence was evaluated using a luminescent plate reader (PerkinElmer 2030). The firefly luminescence was normalized to the Renilla luminescence.

Inhibitor treatment of smooth muscle cells

Inhibitors purchased from Selleckchem were applied at the concentrations recommended by the manufacturer to fibroblasts directly after cell seeding: 10 μM focal adhesion kinase (FAK) inhibitor (PF573228, Cat# S2013), 100 nM c-Src inhibitor (Dasatinib, Cat# S1021). 1:1,000 v/v DMSO was used as vehicle control. qPCR analyses were performed 48 h after treatment.

Western blotting

Aorta segments were sonicated in 200 μL cold RIPA buffer (50 mM Tris pH 8, 150 mM NaCl, 0.5% sodium deoxycholate, 1% Triton X-100, 0.1% sodium dodecyl sulfate), including 1:50 v/v protease inhibitor cocktail (Roche, Cat# 11697498001) and 1:100 v/v phosphatase inhibitor cocktail (Sigma Cat# P5726). 50 μL of the supernatants were loaded onto 12% SDS-PAGE gels. Western blotting was performed with one of the following primary antibodies at a 1:1,000 dilution in 2% non-fat dry milk: anti-CCL2 (Abcam Cat# ab25124), anti-IL-1β (Abcam, Cat# ab9722) and anti-GAPDH (Cell Signaling, Cat# 2118). The blots were imaged using a ChemiDoc imager (Bio-Rad). Band intensities were quantified using ImageJ and normalized to GAPDH staining on the same blot, as described previously [70].

Indirect immunofluorescence

After sequential transcardial perfusion first with PBS and then 2% paraformaldehyde under constant hydrostatic pressure (5–10 mL at 1.2 m elevation, tubing with 3 mm internal diameter), dissected thoracic aortae were fixed in 4% paraformaldehyde for 2 h. Tissues were either directly embedded in paraffin, or in Optimal Cutting Temperature embedding medium (Tissue-Tex, Cat# 4583) after 48 h dehydration in 30% sucrose solution. 5 μm sections were stained with one of the following primary antibodies: Anti-C-terminal half of mouse fibrillin-1 (α-mFbn1-C) [71], anti-MMP12 (Abcam, Cat# ab52897), anti-human tropoelastin [72], anti-CD68 (Abcam, Cat# ab5690), anti-CD3 (Abcam, Cat# ab5690) and anti-HIF-1α (Novus Biologicals, Cat# NB100-479SS). Anti-mouse/rabbit Cy5-conjugated antibody (ThermoFisher) was used as secondary antibody.

For staining of cells, 50,000 cells/well were grown in 8-well chamber glass slides. Anti-C-terminal half of human fibrillin-1 [72] and anti-fibronectin (Abcam, Cat# ab6328) were used as primary antibodies. Secondary antibodies were goat anti-rabbit conjugated to Cy3 (ThermoFisher). Nuclear counterstaining was performed with 4′,6-diamidino-2-phenylindole (DAPI). All images were recorded using an Axio Imager M2 microscope (Zeiss) equipped with an ORCA-flash 4.0 camera (Hamamatsu). Original images in the tiff format were exported for cell counting using ImageJ [73], as previously described [70].

Quantification of the aortic lumen area and media thickness

Cross sections of the pressure-fixed aortae were used to determine the lumen area and media thickness. The lumen area was calculated by measuring the perimeter of the inner elastic laminae (Pi) visualized by autofluorescence at 50 × magnification; lumen area = π × (Pi/2π)². The average thickness of the aortic media was calculated using the perimeter of the outer elastic laminae (Po) and Pi determined from the cross sections; media thickness = (Po – Pi) / 2π.

miR-122 in situ hybridization of aorta

In situ hybridization of miR-122 was performed per manufacturer instruction using miRCURY LNA miRNA ISH Optimization Kits (Qiagen, Cat# 339453). As the miR-122 probe is labeled with digoxigenin, mouse Alexa 647-conjugated anti-digoxigenin secondary antibodies (Jackson Laboratory, Cat# 200-602-156) were used. Stained sections were preserved in vector shield (Vector Laboratories).

Multiplexed immunoassay

Conditioned media without serum were collected after 48 h of human smooth muscle culture treated with either 50 nM miR-122 inhibitor or scrambled control. Levels of 178 secreted proteins in the conditioned medium were measured using the nELISA multiplexed bead-based immunoassay profiling service (nplex biosciences, Montreal, Canada) [74]. Readouts were normalized to the total protein level in the conditioned medium measured by BCA Protein Assay Kit (Thermo scientific, Cat# 23225).

Statistical analyses

All statistical analysis were performed using the OriginPro version 2021 software (OriginLab). Prior to the significance analyses, outliers were determined by the Grubbs’ test at a 95% confidence level. Shapiro–Wilk tests were performed to determine data distribution. For normally distributed data, statistical significances were determined using two-sample t test, paired sample t test or one-way ANOVA at a confidence level of 95%, depending on the design of the experiments. For non-normal distributed data, nonparametric tests were performed: Mann–Whitney U test and Wilcoxon signed rank test were used for independent samples and paired-samples, respectively. All data were expressed as mean values ± standard deviation. In relative analyses, the averages of the control groups were set to 1.

Study approval

All mouse experiments strictly followed the guidelines of the Canadian Council on Animal Care and were approved by the McGill University Animal Care Committee (protocol 2017–7979). Human samples for RNAseq analyses were collected in accordance with New York University (NYU) Langone Medical Center Institutional Review Board policies (protocol i16-01807). Informed consent was obtained from each patient. Written informed consent was received prior to participation.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AAA

Abdominal aortic aneurysm

DMSO

Dimethyl sulfoxide

ECM

Extracellular matrix

FAK

Focal adhesion kinase

FBS

Fetal bovine serum

GSEA

Gene set enrichment analyses

HIF-1α

Hypoxia-inducible factor 1α

MFS

Marfan syndrome

miRNA

MicroRNA

MMP

Matrix metalloproteinase

PBS

Phosphate-buffered saline

PSG

Penicillin–streptomycin–glutamine

TAA

Thoracic aortic aneurysm

TGF-β

Transforming growth factor β

αSMA

α-Smooth muscle actin

Author contributions

Study conception and design: RMZ, DPR. Acquisition of data: RMZ, KT, BR, MLM, SK. Analysis and interpretation of data: RMZ, DPR. Manuscript writing: RMZ, DPR. Manuscript editing: KT, BR, SK, MLM, NEHD.

Funding

This work was supported by the Canadian Institutes of Health Research (PJT-162099), the Marfan Foundation (USA), The Natural Sciences and Engineering Research Council of Canada (RGPIN-06278), the Genetic Aortic Disorders Association Canada, and the China Scholarship Council.

Data availability

The microarray data have been deposited in the NCBI's Gene Expression Omnibus database https://www.ncbi.nlm.nih.gov/geo and are accessible through GEO Series accession number GSE199285. All other data sets of this study will be made available upon request to the corresponding author.

Declarations

Conflict of interest

The authors have no conflicts of interest.